Human Toxicity Potential: A Lifecycle Evaluation in Current and Future Frameworks for Hydrogen-Based and Battery Electric Buses in the European Union

Abstract

In recent years, governments have promoted the shift to low-emission transport systems, with electric and hydrogen vehicles emerging as key alternatives for greener urban mobility. Evaluating zero- or near-zero tailpipe solutions requires a Lifecycle Assessment (LCA) approach, accounting for emissions from energy production, components and vehicle manufacturing. Such studies mainly address Greenhouse Gas (GHG) emissions, while other pollutants are often overlooked. This study compares the Human Toxicity Potential (HTP) of Battery Electric Vehicles (BEVs), Fuel Cell Vehicles (FCVs), Hydrogen Internal Combustion Engine Vehicles (H2ICEVs) and hybrid H2ICEVs for public transport in the European Union. Current and future scenarios (2024, 2030, 2050) are examined, considering evolving energy mixes and manufacturing impacts. Results underline that BEVs are characterized by the highest HTP in 2024, and that this trend is maintained even in future scenarios. As for hydrogen-based powertrains, they show lower HTPs, similar among them. This work underlines that current efforts must be intensified, especially for BEVs, to further limit harmful emissions from the mobility sector.

1. Introduction

The transportation sector is one of the primary contributors to environmental pollution and Greenhouse Gas (GHG) emissions in Europe [1]. According to the European Environment Agency (EEA), in 2022 the transport sector alone accounted for around 29% of the European Union (EU)’s total GHG emissions [2]. The demand for cleaner, more sustainable transport solutions has never been more urgent, driven by the pressing need to mitigate climate change, reduce air pollution and achieve the targets set in international agreements like the Paris Agreement [3]. In this context, urban areas, with their dense concentrations of vehicles, become focal points for the challenge of reducing transportation-related emissions. Cities across Europe are facing severe air quality issues, strictly related to pollutant concentrations [4]. This is critical for human life, as high concentrations of CO could lead to permanent cerebral damage (up to death) [5], while long exposure to NO_X, particulate matter (PM) and ammonia (NH₃) concentrations could lead to breathing and cardiovascular diseases [6,7,8]. Furthermore, SO_X alter the water cycle, increasing acidification of basins and oceans, leading to acid rain and altering biodiversity [9].

Limiting the number of circulating vehicles and enhancing collective transportation solutions have been pointed out as the best solutions to cut down transportation-related emissions [10]. Traditional diesel-powered buses, which have been the backbone of public transport for decades, contribute significantly to urban pollution and CO₂ emissions [11,12]. As European nations strive to meet ambitious emission reduction goals, there is an increasing shift towards developing cleaner powertrain technologies for collective urban transportation systems [13,14].

Alternative propulsion systems have been proposed such as Battery Electric Vehicles (BEVs), which allow us improve air quality through their non-existent tailpipe emissions [15], and new fuels (such as hydrogen, e-fuels and ammonia [16,17,18,19]) have shown the potential to reduce the emissions related to the transport sector [15]. Among the Hydrogen-Fuelled Vehicles (HFVs), three main concepts have emerged [17,20,21]. As a matter of fact, hydrogen can be adopted in Fuel Cell Vehicles (FCVs), where hydrogen is converted into electricity, producing only water vapor as a byproduct [22,23], or as a fuel in Hydrogen Internal Combustion Engine Vehicles (H2ICEVs) [17]. As for FCVs, the most promising solution in the road transport sector is represented by Proton-Exchange Membrane Fuel Cells (PEMFCs), thanks to their relatively low operating temperatures and fast start-up times, even though they require hydrogen of relatively high quality [22,24,25]. Aminudin et al. [26] have performed an overview of pros and cons of several FCVs, highlighting that this kind of vehicle is expected to grow rapidly and be widely embraced by customers. Moreover, the authors assess that FC manufacturing is forecasted to become cheaper, thus enabling mass production and further promoting their widespread integration into the transportation market [26]. One example of a FCV already employed for urban transportation is represented by Toyota SORA in Tokyo, demonstrating the viability of FCVs in dense urban settings, particularly when combined with renewable hydrogen production initiatives like Japan’s Fukushima Hydrogen Energy Research Field [27,28]. Pulvirenti et al. [29] compared a 12 m diesel bus against an HH2ICEV and a FCV version, developing ad hoc energy management strategies to limit aging, demonstrating that HH2ICEVs and FCVs are able to reach, respectively, −29% and −42% energy consumption with respect to their diesel counterpart. On the topic of prolonging the device’s life, Hahn et al. [30] studied FC aging effects on vehicle’s performance. The results show that an efficiency-optimized strategy, with respect to a more lifetime-oriented one, yields an increase in hydrogen consumption of about 1% and a reduction in the degradation penalty of about 13%, enhancing the stack’s lifetime at the cost of lower efficiency [30]. Pitsiavas et al. [31] presented two AI-based energy optimization strategies, aimed at optimizing driving behavior and automating refueling, leading to an increase in range, FC lifetime and efficiency. On further developments of this technology, Bhardwaj et al. [32] studied a novel architecture for FC buses combining FCs, batteries and ultra-capacitors to enhance performance, responsiveness and reliability. Results show that this architecture allows us to achieve better power management, higher system efficiency and prolonged vehicle life. On the other hand, in H2ICEVs, despite hydrogen combustion theoretically only producing water vapor, high combustion temperatures coupled with lean mixtures (Hydrogen Internal Combustion Engines, H2ICEs, are usually run in lean conditions [17]) induce the formation of NO_X emissions [33]. Recently, even series hybrid H2ICEVs (HH2ICEVs) have been studied [20], allowing us to optimize H2ICE management and enhance their efficiency.

Despite a hydrogen-fueled powertrain allowing us to drastically reduce pollutant and GHGs emissions, the overall impact of these technologies is even strongly dependent on the hydrogen production method. According to Incer-Valverde [34], hydrogen can be produced following different pathways, leading to different environmental impact and costs. Mainly, hydrogen can be produced from carbon-based compounds (such as methane) leading to carbon dioxide emissions (gray hydrogen), which can also be captured and separated by means of carbon capture techniques (blue hydrogen). Lastly, hydrogen can be produced by splitting the water molecule into hydrogen and oxygen following an electrolysis process; this method allows us to produce zero GHG emissions if the energy adopted to feed the process is fully renewable (green hydrogen) [34].

However, the introduction of such new technologies (such as electric, which produce zero tailpipe emissions) has underlined the need for a complete Life Cycle Assessment (LCA) of the emissions associated with the vehicle in its entire life, without focusing just on tailpipe emissions [35]. LCA is essential in understanding the true environmental impact of transportation technologies. Lifecycle emissions are not only limited to the emissions produced during the use phase (i.e., combustion emissions and components wear), but also include those generated during vehicle manufacturing, fuel/electricity production and end-of-life disposal. To make informed decisions about the adoption of new powertrain technologies, it is crucial to evaluate the full lifecycle emissions of these collective urban transportation solutions.

Numerous studies have been conducted on equivalent LCA CO₂ emissions [15,36,37,38,39], being that CO₂ represented the main GHG emission, while pollutant emissions have been historically studied only from a “tailpipe-only” point of view. Nonetheless, recently, even LCA on pollutant emissions analyses have been published [40,41]. Among HFVs, only H2ICEVs produce tailpipe emissions [42]. It has been demonstrated that, when properly managed, NO_X emissions of H2ICEs can be limited to 0.05 g_NOX/kWh [20,43].

More recently, a growing concern about non-combustion emissions has emerged. Focusing on direct non-combustion emissions (i.e., emissions directly produced by the vehicle excluding combustion processes), road, tire and brake pad abrasion and wear are the main factors responsible for PM emissions, overcoming the combustion-related ones [44]. Beddows et al. [45] have studied the problem, demonstrating that such emissions are directly linked to vehicle class, road type, driving style and vehicle mass. By broadening the circle to encompass direct emissions from fuel/electricity production, vehicle/powertrain manufacturing and the entire supply chain, it is possible to more accurately quantify the true footprint of transportation [46,47]. Electricity generation’s impact has been widely studied in the literature [48], demonstrating that GHGs and pollutant emissions are strictly related to the energy mix (EM) considered, and the plant’s efficiency and generation. Even hydrogen production, as previously stated, is responsible for different pollutant emissions, based on the production method considered. For instance, Sun et al. [49] studied the environmental impact of gray hydrogen produced by Steam Methane Reforming (SMR), concluding that PM, NO_X, SO_X and CO are directly produced from these production method components (i.e., chassis, engines, electric motors, fuel cells, batteries, H₂ tanks, etc.); raw material extraction, manufacturing and subsequent vehicle assembly also bring about several pollutant emissions. Several works in the literature analyzed and roughly quantified the basic materials required to manufacture the aforementioned components, for several possible types and sizes [21,50,51,52,53,54,55,56,57]. Moreover, the EEA [58] provides reports on the emissions of several chemical species related to the extraction of a multitude of different raw materials. As for the manufacturing processes of such components, specific emissions are generated as well. Szumska [59] quantified the emissions of PM and NO_X related to vehicle and engine production. As regards Fuel Cell (FC) emissions, Simons and Bauer [60] quantified in 0.54 kg_SO2/kW the terrestrial acidification potential with a corresponding PM₁₀ emission of 0.125 kg_PM10/kW. As for batteries, Grant [61] quantified the emissions of a lithium-ion nickel–manganese–cobalt oxide (NMC) battery as 800 g_SOX/kWh, 96.9 g_NOX/kWh and 47.9 g_PM10/kWh. These results underscore the need for a precise pollutant LCA analysis to properly quantify the environmental impact of the different solutions.

However, the environmental impact of the different pollutant species varies significantly, as each type of pollutant can affect human health, ecosystems and climate in different ways, as previously discussed. Therefore, a simple addition of the different species is not representative of the real Human Toxicity Potential (HTP) [62], but equivalence conversion factors are needed to compare different species. HTP enables the comparison of pollutants based on their potential harm to human health, allowing for more effective environmental assessments and help in prioritizing mitigation efforts, guiding policymakers towards more comprehensive and targeted actions to address the full range of pollutants and their long-term effects. During the years, different methods to compare harmful emissions have been developed [62,63,64,65,66]. There is still not a unified standard procedure to assess HTP: Huijbregts [64] suggested to evaluate the equivalent 1,4-dichlorobenzene (1,4-DCB_add), Hertwich et al. [65] proposed to evaluate the equivalent toluene (Toluene_add), while Klimenko [66] suggested calculating the equivalent CO emissions (CO_add), discovering that, in terms of CO_add, Euro 6d+ heavy duty diesel vehicles emit 26.9 g_COadd/km while electric vehicles emit only 6.4 g_COadd/km.

This paper focuses on the analysis of different powertrain concepts for the urban context in terms of consumption and equivalent CO emissions (CO_add) in the current and future energy scenarios, including both electric energy and hydrogen production mixes. Different powertrain layouts have been investigated. With respect to other previous studies, in addition to FCVs and BEVs, H2ICEVs and HH2ICEVs have been included, with hydrogen-fueled vehicles being one of the most promising solutions for such sectors. A typical homologation cycle has been set as a reference, considering different scenarios, varying the curb weight and the number of battery/FC/H2ICE swaps. In addition, an LCA analysis has been carried out to estimate the overall CO_add emissions of these solutions, for 2024, 2030 and 2050 scenarios, based on the current energetic panorama of the 27 members of the EU (EU27) and on achievable and forecast trends regarding future developments and emission reductions.

2. Materials and Methods

2.1. Vehicle and Powertrain Characteristics

To lay the foundations for the LCA analysis, a preliminary step of vehicle and powertrain configuration selection and sizing has been performed. The study focuses on urban buses, considering the same approach taken in [15]: a comparison of four different innovative powertrain layouts (H2ICEV, HH2ICEV, FCV and BEV), evaluating two possible curb weights (9 and 12 tons). The characteristics of each layout have been taken from [15] and are summarized in

Table 1. Powertrain characteristics [15].

		H2ICEV	HH2ICEV	FCV	BEV
H2ICE peak power	[kW]	100	100	-	-
Electric motor peak power	[kW]	-	100	100	100
FC peak power	[kW]	-	-	100	-
Battery energy	[kWh]	-	15	15	550
H2 tank capacity 1	[kg]	(36, 42)	(33, 39)	(27, 32)	-

¹ The sizes reported for each HFV layout refer to the 9- and 12-ton versions, respectively.

[15]. The overall weight of each vehicle was determined as the sum of curb weight, total passenger weight (assuming an average passenger weight of 80 kg [67] and an average passenger load of around 66% of the maximum total capacity of the bus [68,69]), powertrain, battery and H₂ tank weights (assuming typical specific weight ratios [50,70,71]).

Regarding HFVs’ H₂ tanks and BEV’s battery sizing, they have been determined in order to achieve the target daily range of 200 km [72,73], based on the consumptions reported in [15]. As for the overall lifespan of each vehicle, assuming around 60,000 km/year driven [74] and around 17 years of service life [75], the estimated total mileage of each vehicle resulted in around 1 million km.

2.2. Lifecycle Pollutant Evaluation Model

As previously stated, the LCA procedure encompasses the evaluation of all pollutant emissions of the vehicle over its whole life cycle, from raw material extraction, components manufacturing, vehicle assembly and energy/fuel production to end-of-life disposal. Three reference scenarios (2024, 2030 and 2050) have been considered, taking into account the forecasted development of the electricity/hydrogen production markets in the EU27, and also taking into account alternative scenarios in terms of curb weight, powertrain replacements and hydrogen production mixes.

2.2.1. Components, Raw Material Extraction, Manufacturing and Assembly

In the introduction, it was highlighted that raw material extraction and manufacturing of components lead to several pollutant emissions [21,50,51,52,53,54,55,56,57,58].

To account for the emissions of raw material extraction, Equation (1) is provided: it reports the evaluation of the overall mass of the z-th chemical species emitted during the raw material extraction process for the c-th component (m_{RawMatExtr,z,c}).

$$m_{RawMatExtr,z,c}={\displaystyle \sum }_k{W}_c·{\%}_{Mat,k,c}·m_{z\leftarrow k}$$

(1)

In particular, W_c refers to the mass of the c-th component, %_Mat,k,c represents the mass fraction of the k-th raw material needed to manufacture the c-th component, and m_z←k represents the mass of the z-th chemical species that is emitted during the extraction of 1 kg of the k-th raw material. By combining these pieces of information, considering suitable conversion factors [66], the HTP contribution of the raw materials extraction phase could be evaluated.

As for the vehicle assembly and components manufacturing processes, these contributions have been evaluated following Equations (2)–(4).

$$m_{Ass,z,i}=W_i·m_{Ass,z}$$

(2)

$$m_{Manuf,z,c}=P_c·m_{z,c}$$

(3)

$$m_{Manuf,Batt,z,i}=C_{Batt,i}·m_{z,Batt}$$

(4)

In Equation (2), m_Ass,z,i represents the overall mass of the z-th chemical species emitted during the assembly process of the i-th layout (i.e., H2ICEV, HH2ICEV, FCV, BEV), W_i represents the overall weight of the i-th layout, and m_Ass,z represents the mass of the z-th chemical species that is emitted during the assembly process, per kg of vehicle. In Equation (3), m_Manuf,z,c represents the overall mass of the z-th chemical species emitted during the manufacturing process of the c-th component (i.e., H2ICE, FC, electric motor), P_c represents the power rating of the c-th component, and m_z,c represents the mass of the z-th chemical species that is emitted during the manufacturing process of the c-th component, per kW of said component. Finally, in Equation (4), m_{Manuf,Batt,z,i} represents the overall mass of the z-th chemical species emitted during the manufacturing process of the battery of the i-th layout, C_Batt,i represents the battery capacity of the i-th layout, and m_z,Batt represents the mass of the z-th chemical species that is emitted during the manufacturing process of the battery, per kWh.

Szumska [59] analyzed the lifecycle pollutant emissions related to vehicle and engine production. It was reported that, in terms of vehicle assembly, the process leads to 1.39 g_NOX/kg_vehicle and 2.22 g_PM/kg_vehicle [59]. As for internal combustion engines, the estimated lifecycle pollutant emissions are around 80 g_NOX/kW_engine and 30 g_PM/kW_engine for NO_X and PM, respectively [59].

In relation to electric motors, de Souza et al. [76] investigated the lifecycle environmental impacts of different kinds of these devices. In particular, the authors have determined that, during the manufacturing phase of permanent magnet synchronous motors (PMSMs), around 349.1 g_SOX/kW_{ElectricMotor} and 62.4 g_PM/kW_{ElectricMotor} of SO_X and PM, respectively, are generated [76].

Moving to FCs, Simons and Bauer [60] focused on LCA impact of PEMFC systems (stacks and Balance of Plant, BoP, components) for automotive applications. In particular, they determined that the production process of the whole FC system generates around 540 g_SO2/kW_FC and 130 g_PM/kW_FC with respect to the system’s peak net power, with the Membrane-Electrode Assembly (MEA) being the main responsible, representing around 90% and 80% of the aforementioned values, respectively [60].

Finally, as for batteries, different Lithium-ion chemistries are employed in automotive applications. The main ones are NMC, Nickel–Cobalt–Aluminum Oxide (NCA), Lithium Iron Phosphate (LFP), and Lithium Manganese Oxide (LMO) [55,56,57,77]. Figure 1a reports a comparison of the CO_add emissions of these different technologies per kWh, including both raw material extraction and manufacturing in 2024. As it can be seen, the emissions are similar among the different chemistries and range between 49 and 52 kg_COadd/kWh. Despite being the worst ones in terms of emissions, NMCs were still chosen for this analysis, being the most adopted technology in the automotive sector [77]. Grant et al. [61] investigated the environmental impact of lithium-ion batteries’ supply chain. Results show that for the considered NMC batteries, the production leads to 800 g_SOX/kWh, 96.9 g_NOX/kWh and 47.9 g_PM10/kWh [61]. Figure 1b illustrates the CO_add emissions of an NMC battery per kWh, including both raw material extraction and manufacturing over the three reference years of the analysis.

All the aforementioned values (summarized in

Table 2. Pollutant emissions related to the various components [59,60,61,76].

Component		Pollutant
Vehicle assembly	[gNOX/kgvehicle]	1.39	[59]
	[gPM/kgvehicle]	2.22	[59]
ICE	[gNOX/kWengine]	80	[59]
	[gPM/kWengine]	30	[59]
Electric motor	[gSOX/kWElectricMotor]	349.1	[76]
	[gPM/kWElectricMotor]	62.4	[76]
FC	[gSO2/kWFC]	540	[60]
	[gPM/kWFC]	130	[60]
Battery	[gSOX/kWh]	800	[61]
	[gNOX/kWh]	96.9	[61]
	[gPM10/kWh]	47.9	[61]

) have been scaled for the specific case study, based on the different service life, overall vehicle weights and powertrain characteristics (Table 1) considered in this paper. Moreover, for future scenarios, foreseeable emissions reductions have been considered [15,78,79,80,81,82,83,84,85,86].

2.2.2. Energy Production and Consumption

The specific generation method employed to produce the energy source (i.e., electricity or hydrogen) heavily affects the kind and the mass of pollutant emissions yielded [49,87]. Thus, it is quite clear that these kinds of emissions are tightly correlated with each country’s electricity and hydrogen production mixes [88]. As for the 2024 scenario, EU27 electricity [88,89,90] and hydrogen [91] production mixes have been directly taken as reference. On the other hand, for 2030 and 2050 future scenarios, several sources from the literature have been examined [2,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124], considering foreseeable emissions reductions from the generation sector owing to limited infrastructure evolution [124,125]. Regarding future developments of the energy mixes, the European Environment Agency (EEA) identifies three types of projection scenarios: Without Measures (WOM), With Existing Measures (WEM) and With Additional Measures (WAM) [2]. The WOM scenario disregards all policies adopted after the baseline year, projecting future trends solely from historical data [2]. The WEM scenario, instead, includes policies and measures that are already implemented or formally adopted, such as existing legislation, voluntary agreements, or allocated resources. Since it reflects the impact of actions that governments have committed to, WEM is generally considered the most realistic outlook [2]. The WAM scenario goes a step further by incorporating proposed policies that are under discussion and could plausibly be adopted, offering an indication of the additional reductions achievable through future decisions [2]. In some literature sources, in addition to the WEM scenario, WAM paths have been analyzed, with changes from the reference WEM projection, considering plausible alternatives (i.e., different renewable energy sources and nuclear power levels of penetration, socio-economic aspects, import/export impacts, etc.). However, in order to try to consider the most probable scenario without eliminating the contribution of already agreed-upon policies and measures, as it would be in the WOM case, but at the same time avoiding possible excessively ambitious targets of WAM perspectives, this study focuses on WEM projections for the energy mixes. Furthermore, even a 100% green hydrogen scenario (100 GH) has been investigated, evaluating the possibility of having H₂ production and refueling directly at the bus depot.

As far as electricity production is concerned, among all data presented, the 2023 emissions reductions progress report by the United States Environmental Protection Agency [126] provides information on specific emission factors (EFs) of the main pollutant emitted from fossil fuel-based electricity generation, being NO_X and SO_X. Starting from NO_X, it is stated that 0.624 g_NOX/kWh are generated from coal-based plants, 0.1004 g_NOX/kWh from gas-based ones and 0.8163 g_NOX/kWh from oil-based ones. Moving to SO_X, the EFs are 0.86 g_SOX/kWh, 0.0038 g_SOX/kWh and 0.3604 g_SOX/kWh, from coal, gas and oil plants, respectively [126].

In terms of hydrogen production, as already discussed in the introduction, several methods are available. In the context of this work, only gray, blue and green hydrogen have been considered. Being that this study does not account for emissions related to manufacturing of the production plant and the associated facilities, and that the focus is placed on pollutant emissions, gray and blue hydrogen’s impacts are the only effective ones [49]. Sun et al. [49] quantified the pollutant and GHG EFs of SMR facilities in the United States. In particular, they reported that hydrogen produced from SMR accounts for 0.0079 g_CO/kWh_H2 of CO, 0.072 g_PM/kWh_H2 of PM, 0.0227 g_NOX/kWh_H2 of NO_X and 0.0004 g_SOX/kWh_H2 of SO_X [49].

The overall EFs of each pollutant for both electricity and hydrogen production are then calculated, taking into account each EU27 production mix. In particular, in order to account for EF variations during the service life of the vehicles, starting from the reference year selected (2024, 2030 or 2050), the end-of-life (EoL) EFs have been estimated, and average EFs between reference year and EoL have then been considered. Figure 2 shows the CO emissions associated with the electricity generation mix of each EU27 country member over the three reference years. While in most cases emissions are reduced over the years, some countries show counter-intuitive trends. The reason behind this phenomenon is related to the fact that a strong decarbonization of future electricity generation mixes is expected. Unfortunately, this does not always allow countries to completely fulfill the internal energy demand with domestic production. Subsequently, energy import is required, and it usually comes from countries who rely more on fossil fuels to produce electricity, thus worsening the CO_add emissions of the original country. Moreover, future trends for Luxembourg and Malta have not been included due to a lack of data available on future forecasts.

Furthermore, the contribution of transportation of both electricity and hydrogen has been accounted for [39,127,128,129,130]. Starting from the electricity distribution network, the main terms of losses (thus emissions) in the distribution phase of electricity are represented by the high-voltage grid network and the transformation losses from DC/DC conversion [127,128,129]. These losses range from 4% to 29%. Instead of a country-to-country variability of this aspect, in this work a mean value of 14% has been considered, being that huge variability has been reported in the historical data over the years [127,128,129]. The inefficiency of the grid basically gives the extra amount of electricity (thus emissions) to be generated to effectively deliver the requested amount. This has been accounted for as shown in Equation (5):

$$m_{Distrib,EE,z,s}={\displaystyle \sum }_f{\%}_{EE,f,s}·m_{EE,z,avg}·E{E}_{Cons,s}·\left.\left({\displaystyle \frac{1}{{\eta }_{Transf}·{\eta }_{Distrib}}}\right.-1\right)$$

(5)

Equation (5) shows the evaluation of the overall mass of the z-th chemical species emitted due to electricity distribution (m_{Distrib,EE,z,s}) in the s-th EU27 country by the BEV layout. In particular, %_EE,f,s represents the amount of electricity which is produced through non-renewable technologies in the s-th EU27 country, m_EE,z,avg is the mass of the z-th chemical species that is emitted for the generation of 1 kWh of electricity, averaged over the vehicle lifecycle, EE_Cons,s is the consumption, in kWh, of the BEV in the s-th EU27 country, and finally, η_transf and η_distrib represent, respectively, the average transformation and distribution efficiencies.

Moving to hydrogen distribution, there are two main alternatives: either compressed gas or liquid (solid storage in metal hydrides is still under development). Being gaseous, hydrogen is characterized by a very low density at ambient pressure and temperature; a compression stage has to be accounted for to reach the typical storage pressures to be used on-board, requiring around 2.5 kWh/kg_H2, compressed to reach 700 bar [39]. Since liquid hydrogen is particularly expensive for production and must be kept at extremely low temperatures to avoid boil-off, only gaseous hydrogen has been considered. As regards gaseous hydrogen, it can be distributed either via truck trailers in pressurized vessels or via dedicated pipelines [130].

Figure 3 shows a comparison of CO_add emissions per kg_H2 distributed among the EU27 between different possible distribution pathways. In particular, the pipeline solution is compared against three different truck trailer scenarios, with varying distances between the hydrogen production plant and the refueling station (the distance reported is the round trip). The distances have been chosen considering average values from the literature [131,132,133]. The error bars, on the other hand, represent the possible variation in the emissions due to the specific electricity mixes and their different EFs among the EU27 for the hydrogen compression step. Despite being the best performing solution in terms of pollutant emissions (between 9% and 25% of the truck trailer values), the pipeline network for hydrogen distribution is not widespread enough yet. For this reason, in this work, the truck trailer option has been chosen. As for the distance, 200 km has been set (meaning 400 km for the round trip) as an average condition [131,132,133]. In addition, as previously stated, for all 100 GH scenarios, hydrogen has been considered to be produced at the depot. Thus, in those scenarios, any distribution pathway was envisioned, making compression the only contribution considered. Furthermore, it was assumed that the pipeline network would be developed by 2050, thus in all 2050 hydrogen mix (HM) scenarios the pipeline substitutes the truck trailers for H₂ distribution.

$$m_{Compr,H_2,z,s,i}={\displaystyle \sum }_f{\%}_{EE,f,s}·m_{H_2,z,avg}·H_{2,Cons,s,i}·kW{h}_{H_2,Compr}$$

(6)

$$m_{Distrib,H_2,k,z,s,i}={\displaystyle \sum }_f{m}_{H_2,Distrib,k,z,avg}·H_{2,Cons,s,i}·d_{Distrib,k}$$

(7)

Equations (6) and (7) show the evaluation of the overall masses of the z-th chemical species emitted due to H₂ compression (m_{Compr,H2,z,s,i}) and distribution (m_{Distrib,H2,k,z,s,i}) in the s-th EU27 country by the i-th HFV layout. In particular, %_H2,f,s represents the amount of H₂ which is produced through non-renewable technologies in the s-th EU27 country, m_H2,z,avg and m_{H2,Distrib,k,z,avg} are the masses of the z-th chemical species that are emitted for each kg of H₂, that is, respectively, produced or distributed through the k-th H₂ distribution pathway, averaged over the vehicle lifecycle, H_2,Cons,s,i is the consumption, in kg_H2 of the i-th HFV in the s-th EU27 country, kWh_H2,Compr is the energy required to compress 1 kg of H₂ up to 700 bar, and d_Distrib,k is the coefficient related to the distribution round-trip distance required for the k-th H₂ distribution method.

Lastly, to enable the assessment of each vehicle’s use-phase lifetime energy consumption impact, consumption data have been taken from [15] (where for each possible powertrain and curb weight combination, electricity/hydrogen consumptions per km have been determined). The calculations have been carried out for each of the EU27 members, considering each country’s average external temperature (affecting heating, ventilation and air conditioning, HVAC, system’s impact on overall energy consumption) and the overall vehicle weight.

$$m_{Cons,EE,z,s}={\displaystyle \sum }_f{\%}_{EE,f,s}·m_{EE,z,avg}·E{E}_{Cons,s}$$

(8)

$$m_{Cons,H_2,z,s,i}={\displaystyle \sum }_f{\%}_{H_2,f,s}·m_{H_2,z,avg}·H_{2,Cons,s,i}$$

(9)

Equations (8) and (9) show the evaluation of the overall masses of the z-th chemical species emitted due to electricity (m_Cons,EE,z,s, for the BEV layout) or H₂ (m_{Cons,H2,z,s,i}, for the i-th HFV layout) consumption in the s-th EU27 country.

2.2.3. Tire and Brake Wear, and Road Abrasion

The literature suggests that, in modern vehicles, road abrasion, along with brake pads and disk wear, constitutes the dominant source of PM emissions [44]. In particular, Beddows and Harrison [45] analyzed non-exhaust PM emissions for road vehicles. Their analysis suggests that the overall PM₁₀ and PM_2.5 emissions from tire, brake and road contributions depend on the type of road (i.e., urban, rural or motorway) and on the overall vehicle weight, as represented in Equation (10) [45]:

$$P{M}_{Wear,x,i}={\displaystyle \sum }_k{b}_{k,x}·{\left.\left({\displaystyle \frac{W_i}{1000}}\right.\right)}^{{\displaystyle \frac{1}{c_{k,x}}}}$$

(10)

where x represents the PM particle size in μm (i.e., 2.5 or 10), i refers to the specific layout (i.e., H2ICEV, HH2ICEV, FCV, BEV), W_i represents the overall weight of the i-th layout in kg, k refers to the specific PM source (i.e., road, brake or tire) and b and c are characteristic parameters, dependent on vehicle class, type of road and type of emission. Since Beddows and Harrison’s approach is usually employed for ordinary vehicles, while bus paths are usually characterized by higher loads and frequent starts and stops that may affect wear emissions, the modified approach proposed by Tivey et al. [134] has been adopted.

2.2.4. Combustion

In the context of hydrogen combustion, NO_X emissions are expected [135,136,137]. Even PM and CO₂ could be theoretically emitted due to lubricant oil oxidation. However, these emissions are negligible and therefore have not been accounted for in this work. Arsie et al. [20] demonstrated that, when H2ICEs are properly controlled, NO_X emissions in H2ICEs could be limited to just 0.05 g_NOX/kWh. As it pertains to this study, such value has been kept as reference in order to evaluate overall H2ICEs-based powertrain NO_X emissions. Knowing energy consumptions over the reference cycle and the expected service life of the vehicles, total NO_X emissions from combustion were determined for both H2ICEVs and HH2ICEVs, considering the mean engine working point over the reference drive cycle.

2.2.5. End-of-Life Phase

To completely fulfill the LCA analysis, evaluations on the end-of-life impact on all of the emissions should be performed. However, the literature lacks sufficient data to properly complete the estimation, and only partial information for some of the vehicles considered in this work is available [138,139,140,141]. Therefore, to avoid introducing inaccuracies in the results, this section only serves the purpose of providing some insights on the topic.

As regards FCVs, FC recyclability is still under study [138], and no data is reported on the specific pollutant emissions related to this process. However, Chen et al. [138] evaluated the electricity consumption for the recovery of some materials (mainly aluminum, carbon fiber, iron, copper and steel) for a single FCV, resulting in up to 8.83 kWh/vehicle of electricity consumed. Thus, emissions linked to electricity generation can be considered: focusing on the mean electricity mix, this would lead to 5 g_COadd/vehicle.

As regards batteries, there are three main methods for their recycling, namely direct recycle, hydrometallurgy and pyrometallurgy [139]. Pollutant emissions are strictly related to the country where the recycling takes place and to the transport method used [139]. Focusing on pyrometallurgy, which is the most widespread one, the emissions that can be accounted for are quasi-zero in terms of SO_X and up to 5 g_NOX/kg_battery in terms of NO_X, if the recycling takes place back in China, while it becomes less than half in the case of local recycling [139]. Taking this into account, it would increase the emissions by almost 3 g_COadd/km for the BEV and by less than 1 g_COadd/km for the HH2ICEV/FCV. Moreover, recycling emissions at an industrial level have been recorded, with PM10 concentrations of 140 μg_PM10/m³, which is mainly composed of Al, Fe, Cu, Mn and Ni [140].

Moving to the electric motors, the energy consumption for their recycling can be estimated as 66 kWh/ton, considering shredding and the further separation steps, including magnetic separation. Unfortunately, no information is provided on associated pollutant emissions [140]. The only values that are possible to estimate are the equivalent pollutant emissions associated with electricity consumption. This conversion factor is a function of the energy mix considered. For instance, if we consider the electricity generation mixes of Sweden and Poland (being the most different) and the mean EU27, the equivalent emissions associated with the recycling of the electric motor would span between 250 g_COadd/ton, 1940 g_COadd/ton and 831 g_COadd/ton, respectively.

Finally, for combustion engine recycling, nowadays usually the engines are remanufactured with partial substitution of components and rectification [141]. Smith and Keoleian [141] underline that up to −66% of pollutant emissions (specifically CO, NO_X and SO_X) can be reached using a remanufactured engine with respect to a new one.

2.2.6. Equivalency Determination

As mentioned in the introduction, to quantify the overall environmental impact of a vehicle, its overall pollutant emissions must be compared. For this purpose, restricting the analysis to a mere comparison of the total emissions in terms of mass is a flawed approach for quantifying the overall impact of a vehicle, as each chemical species has distinct effects. To ensure a proper and correct comparison, all quantities should be normalized to reference pollutants through appropriate equivalence coefficients.

As previously discussed in the introduction, several approaches for this evaluation have been proposed in the literature [62,63,64,65,66], and for the present work, the one recommended by Klimenko [66] to determine the equivalent HTP has been followed.

In order to determine the equivalence, the approach defined in Equation (11) has been followed, where R_t,CO,z represents the relative toxicity, reported in

Table 3. Relative toxicities for human toxicity substances with respect to CO [66].

Human Toxicity Substance	Rt,CO,z [kg/kg]
NOX	75
SOX	25
NH3	75
PM	200
As	2000
Cd	2000
Cr	1000
Cu	40
Fe	75
Hg	4000
Ni	4000
Pb	400
Zn	40

, of the z-th human toxicity substance with respect to CO_add, and m_z indicates the mass of the z-th toxic substance.

$$C{O}_{add}={\displaystyle \sum }_z{R}_{t,CO,z}·m_z$$

(11)

3. Results and Discussion

The HTP assessment was carried out across all EU27 countries considering three reference years (i.e., 2024, 2030 and 2050). For each reference year, different scenarios have been evaluated, as reported in

Table 4. Scenarios evaluated in each of the three refence years.

Scenario	Curb Weight	Powertrain Replacements	Hydrogen Production Mix
S1 (Baseline)	9 ton	Single	HM
S2	9 ton	Single	100 GH
S3	9 ton	Double	HM
S4	12 ton	Single	HM
S5	12 ton	Single	100 GH
S6	9 ton	Double	100 GH
S7	12 ton	Double	HM
S8	12 ton	Double	100 GH

. As for the hydrogen production mixes, two possible alternatives have been compared: HM and 100 GH. By varying the curb weight (9 tons and 12 tons) and the number of powertrain replacements over the vehicle’s lifetime (single or double replacement over the vehicle’s expected lifetime), sensitivity to these aspects was also analyzed. Focusing on the rationale behind the choice of the powertrain replacement rates, despite the fact that the replacements’ frequency can vary among the different powertrain layouts as a function of operating conditions, it was assumed that, for any given scenario, all powertrain replacements occur the same number of times and at the same moments over the vehicle’s lifetime. As for the specific rates chosen, regarding the battery, an expected life of 2000 cycles can be taken as a reference [142]. Taking into account one recharge/day, around 60,000 km/year driven [41] with a mean of 200 km/day [72,73], the battery lasts between 400,000 and 600,000 km, depending on working conditions (i.e., external temperature and passengers’ load, thus consumptions, depth of discharge and discharge rate). For FCs and H2ICEs an estimated lifetime of 20,000 h [143] can be taken as a reference. Considering the mean speed of the cycle (22.5 km/h), an equivalent life of around 450,000 km can be calculated, which, with a vehicle lifetime mileage of 1 million km, justifies the choice on the powertrain replacement rates investigated. Additionally, to properly evaluate energy consumption, the typical average external temperature of each country and passenger load were taken into account [15,68]. As a matter of fact, both influence overall consumption due to variation in power consumption from HVAC and, for passengers, due to variation in overall weight. A passenger load factor of 66% was considered, following the approached discussed in the reference paper for the design and consumptions estimations used in this work [15], which was based on a literature review [68,69]. Moreover, in order to properly highlight the impact of these novel solutions with respect to the current most common solution, which is represented by diesel buses, the CO_add emissions of this kind of powertrain have been estimated, following the analysis reported in [144]. Considering the same conditions of the baseline scenario of this study, the results showed that the diesel architecture generates 153.6 g_COadd/km.

3.1. Input Analysis

3.1.1. Input Uncertainty Analysis

A preliminary uncertainty analysis evaluation was carried out on the effect of some possible alternatives to the final configuration in the current framework (i.e., 2024). The uncertainty analysis scenarios are reported in

Table 5. Uncertainty analysis alternatives evaluated against the 2024 baseline scenario (S1).

Scenario	Battery Type	H2 Distribution Pathway	Truck Trailer Round-Trip Distance
S1 (Baseline)	NMC	Truck trailer	400 km
UA1	NCA	Truck trailer	400 km
UA2	LFP	Truck trailer	400 km
UA3	LMO	Truck trailer	400 km
UA4	NMC	Truck trailer	200 km
UA5	NMC	Truck trailer	600 km
UA6	NMC	Pipeline	-

. The considered baseline scenario (S1) is the same as the one described by Table 4, with the additional explicit identification of the battery type (NMC), the H₂ distribution pathway (truck trailer) and the round-trip distance that the truck trailer needs to cover (400 km). As for the other alternatives, it was decided to analyze the effect of changing only one of the previously mentioned categories each time with respect to the baseline scenario.

Figure 4 shows the results of this preliminary evaluation, where the error bars represent the minimum and maximum values reached among the EU27 countries, mainly due to country-to-country variability in electricity and H₂ production mixes, as well as consumptions, as stated in the beginning of this section. As can be clearly seen, the first three alternative scenarios (i.e., from UA1 to UA3), where only the battery technology changes, affect only BEVs, and the variation ranges between −3% for NCA and −6% for LFP and LMO. This is due to the fact that, as illustrated in Figure 2, the difference in emissions among the different chemistries is relatively limited, meaning that the overall variation becomes appreciable only for BEVs, where the battery capacity is more than thirty times that of the other HFVs that employ a battery (i.e., HH2ICEV and FCV).

On the other hand, moving to the scenarios where the only changes are represented by a specific hydrogen distribution pathway and round-trip distance that the eventual truck trailer should cover (i.e., from UA4 to UA6), it is self-evident that only HFVs are influenced. In particular, UA4 and UA5 are characterized by a −50% and +50% variation in the truck trailer round-trip distance with respect to the baseline scenario. Consequently, the resulting shift is symmetrical for all HFVs, being ±16% for H2ICEVs, ±13% for HH2ICEVs and ±11% for FCVs. Finally, in UA6, the assumed distribution method changes completely, going from truck trailers to pipelines, thus allowing for a strong reduction in the overall emissions for all HFVs (from −23% for FCVs up to −33% for H2ICEVs). The largest reduction from UA4 to UA6 is always shown by H2ICEVs because of their higher overall H₂ consumption with respect to the other HFVs, meaning that a larger amount of hydrogen to be distributed is required, thus making the distribution itself more impactful.

3.1.2. Input Data Limitations

As far as limitations on input data are concerned, future energy mix projections represent a crucial source of possible uncertainty. As previously stated in Section 2.2.2, the EEA distinguishes between three types of future projection scenarios: WOM, which are purely based on historical trends, neglecting any kind of already adopted measures, WEM, which encompass all policies that governments and stakeholders have already been implementing, thus becoming the most realistic futures, and WAM, which, with respect to the previous ones, include additional actions under discussion that are plausible to be enacted [2]. The National Energy and Climate Plans (NECPs) [145] introduced by the EU outline how each EU27 country intends to address several environment-related topics such as decarbonization, energy efficiency and security, energy market and research and innovation, both in the short (up to 2030) and in the long terms (up to 2050), in order to achieve the goals set by the Green Deal [146]. The objectives have been developed both on the single country’s level and on the EU27 level as a whole, in order to promote collaboration between governments, citizens and businesses [145]. For this reason, we have examined several literature sources [92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124] to identify the forecasted EU-27 energy mixes in 2030 and 2050 WEM scenarios.

In some literature sources, in addition to the WEM scenario, WAM paths have been analyzed, considering plausible alternatives, with changes from the reference WEM projection. These deviations mainly represent greener and more ambitious futures, but some of them also envision even worse cases (i.e., Renewable Energy Sources, RESs, limited availability, weak adoption of RESs). The largest amount of developed WAM alternatives revolve around a variation in RES penetration into each country’s energy market, mainly evaluating the consequences of increasing the energy generated from RESs [92,93,98,99,100,103,105,115,116,118,121,122]. Among the various RESs, the ones whose impacts were mainly investigated were solar [103,105,115] and wind [103,105,117], with some sources even taking into consideration the option of relying mainly on biomasses [100,103]. As it could be easily understood, a higher adoption of RESs would lead to a much lower emission factor for electricity production and would also have a positive effect on the economy in the long term, but it would require really high investments [98,100,105] in the short term and higher technology development to be able to fulfill the demand [99,103,123]. These analyses comprised not only technological aspects, but also social impacts, such as opposition to geological occupation by onshore wind and solar plants, easiness of adoption of RES technologies and cooperation between countries. In particular, some sources evaluated alternative projections where an opposition to RESs was present [100,115,117,120], highlighting, for example, the necessity of favoring offshore plants to avoid excessive impact on landscaping [115,117]. Another option to compensate for unrealistic RES share scenarios is the possibility of phasing out completely coal-fired plants by substituting their production with gas-powered plants coupled with carbon capture and storage systems [98,100,103]. This would be the easiest solution in terms of feasibility but would not allow us to comply with emissions reduction targets and would have a negative impact on the EU economy in the long term. As far as the economy is concerned, some authors even performed sensitivity analyses on price variations of CO₂ storage, natural gas extraction, batteries, solar panels and wind turbine production to evaluate their impact on RES penetration capability [107,113], concluding that the lower the prices, the easier the transition to these solutions would be. Another variable used to define alternative WAM paths is represented by the influence of nuclear energy [100,103,105,122]. These evaluations were mainly limited to those countries where nuclear energy is already employed, in order to analyze possible variations in its share. The main trend was to envision future scenarios where nuclear energy was limited or completely phased out, in favor of relying more on RESs. Regardless, it was highlighted that the exploitation of nuclear energy in place of fossil fuels could help in fulfilling energy demand and in lowering overall emission factor for electricity production [103,105,122]. Interconnectivity between countries and power transmission is an additional aspect that some sources have taken into consideration in their analyses [95,100,103,104,116]. The shared opinion is that, when pushing for greener electricity generation through non-programmable RESs (i.e., wind and solar), it is fundamental to improve networks for cross-border transmission in order to be able to possibly share over-production and avoid energy waste. Moreover, in an idealistic forced condition of 100% production from RESs, it would be improbable for countries with larger energy demands to completely fulfill the request with only domestic generation, making importing from other countries a necessity. On the other hand, importing usually occurs at a higher EF than that of the importer’s country, with an inevitable detrimental effect on the importer country’s emission factor for electricity production [95].

As previously stated, to keep a realistic perspective, without neglecting measures that countries already committed to, as in WOM scenarios, and at the same time avoiding excessively optimistic WAM assumptions, WEM projections have been taken as reference.

Moving to manufacturing emission factors, the main development concern FCs and batteries, being that ICEs are already well established as a technology [82,83,84,85,86]. Starting from FCVs, H₂ tanks and Pt loading represent the most impactful aspects of the manufacturing processes of FC systems [82,83]. Benitez et al. [82] report that expected decarbonization of electricity mix coupled with more efficient carbon fiber production processes and a lower amount of composite material should lead to around −50% emissions by 2050 from the manufacturing of H₂ tanks. Evangelisti et al. [83] analyzed the LCA of a FCV, focusing on the manufacturing phase, showing that, if forecasted 2030 reductions in Pt loading for FC stacks are achieved, it would lead to a −25% reduction in emissions from the manufacturing phase. Moving to batteries, the identified criticalities are related to the electricity mix and the manufacturing processes efficiency [84,85,86]. The Swedish Environmental Research Institute [84] carried out an investigation on the emissions related to NMC battery manufacturing, showing that future trends for materials shares, process efficiencies and electricity mix decarbonization should lead to a noticeable reduction in emissions. These results have been confirmed by other studies on the matter [85]. Moreover, Peiseler et al. performed a review of possible strategies to reduce batteries’ carbon footprint, underlining the importance of improving manufacturing process efficiency and promoting the decarbonization of the electricity mixes.

3.2. Current Scenario

Figure 5 reports the average CO_add emissions of the four layouts in the EU27 in 2024, comparing the several scenarios previously defined. Also in this case, the error bars represent the minimum and maximum values reached among the EU27 countries. The first scenario (S1) has been selected as the baseline, and all other scenarios show the variation, in percentage, with respect to it, for each layout.

As can be clearly seen, regardless of the type of powertrain or of the specific scenario considered, all the options report much lower CO_add emissions with respect to the diesel case, ranging from 87% down to even 17% of diesel’s values. Furthermore, it is self-evident that, in all the scenarios that include 100 GH, BEVs’ CO_add emissions do not change with respect to the corresponding scenarios where HM is the only difference, being that BEVs are not affected by the hydrogen production mix. Moreover, 100 GH in all EU27 countries removes the country-to-country variability of the hydrogen production mix, making the error bars almost coincide with the average values for all the HFVs in those scenarios (a difference is still present, due to different consumptions). In addition, 100 GH scenarios lead to a noticeable drop in emissions for these kinds of layouts. BEVs are characterized by the highest CO_add emissions overall, being between two to almost four times those of HFVs in all scenarios, while HFVs show very similar values amongst them, with H2ICEVs being the lowest. As for the sensitivity to other parameters, considering only those scenarios where just one parameter is changed with respect to the baseline (i.e., S2, S3 and S4), BEVs and FCVs are more negatively affected by doubling the powertrain replacement, with BEVs being the one with the largest impact, (+34% for BEVs and +19% for FCV) with respect to increasing the curb weight (+5% and +17%, respectively). On the other hand, H2ICEVs and HH2ICEVs show the opposite trend, reporting higher CO_add values in the S4 scenario rather than in S3. These different behaviors could be traced back to the fact that the raw material extraction and manufacturing phases of batteries and FCs are noticeably more impactful than the respective phases for H2ICEs. Conversely, being more efficient, and considering the same curb weight, the consumptions and thus the emissions of BEVs and FCVs related to the in-use phase are much lower than those of H2ICEVs and HH2ICEVs. Despite having a battery as well, FCV and HH2ICEV CO_add emissions are far from those of BEVs due to the large difference in battery capacity.

To ensure the robustness of the results, a comparison with Toluene_add emissions values has been performed (Figure 6), following the approach proposed by Hertwich et al. [65]. As can be clearly seen, this approach shows similar trends to the one based on CO_add proposed by Klimenko [66], widening even more the gap between BEVs and HFVs. Since the CO_add approach is the more conservative one, it has been chosen as the only one to investigate for the other future scenarios of this work.

3.3. Future Scenarios

Future scenarios for 2030 and 2050 have been investigated, taking into account the expected evolution of electricity and hydrogen production mixes in all EU27 countries, together with foreseeable reductions in CO_add emissions related to the manufacturing process [15].

3.3.1. 2030 Scenario

Some general remarks already made for 2024 can be extended to the 2030 scenarios (Figure 7) as well. In particular, BEVs are still the worst-performing layout amongst the four. CO_add emissions see a reduction in all scenarios, with the largest pertaining to BEVs. Due to the progressive decarbonization of the electricity and hydrogen production mixes over the whole EU, the width of the error bar variability range is more limited for all powertrains. Despite this transition to greener technologies, the effect of distribution is still very impactful. For this reason, limited differences can be identified between 2024 and 2030 scenarios in relative terms.

3.3.2. 2050 Scenario

The 2050 scenarios are reported in Figure 8. Also in this case, the gradually greener production mixes reduce the country-to-country variability, and thus the gap between the error bars. Moreover, the CO_add emissions decrease, with BEVs reaching values that are around half of BEVs’ values in 2024, but still around two times those of HFVs. The expected reduction in battery-related emissions (−50% with respect to 2024) heavily influences BEV emissions in scenarios where a double replacement is considered, slightly reducing its worsening effect (from +34% in 2030 to +33%). On the other hand, FC manufacturing-related reduction is more limited (−30% with respect to 2024), thus the improvement is marginal. From the forecasts, hydrogen production in 2050 should be almost fully green, thus tending towards closing the gap between HM and 100 GH. This translates to baseline scenarios for HFVs where the hydrogen mix effect is already similar to the one in 100 GH scenarios. This aspect is also supported by the assumption of having an already developed pipeline distribution network for H₂, which contributes to reducing overall emissions. For these reasons, 100 GH scenarios in 2050 no longer provide large improvements as it was for 2024 and 2030 (between −7% and −12% in 2050 S2 with respect to between −34% and −49% in 2030 S2), and this positive effect is more than compensated by the detrimental impact of larger curb weights and more powertrain replacements in scenarios where these effects are combined (i.e., S5, S6, S7 and S8).

3.3.3. Baseline Scenario Development and Comparison

Finally, Figure 9 shows the development of the CO_add emissions of the four layouts over the three refence years in the baseline scenarios, in Italy, Poland and Sweden, explicitly distinguishing between the different contributions considered: Wear (of brakes, tires and road abrasion), Fuel/Energy utilization (i.e., consumption and distribution), Combustion (only for H2ICEs-based layouts), Assembly of the vehicle, Fuel/Energy Storage (i.e., batteries and H₂ tanks) and Powertrain (i.e., H2ICE, EM, FC). These countries have been selected due to their diverse electricity and hydrogen mixes, which affect the overall emissions. Being that the scenario considered is always the same (curb weight of 9 ton, single powertrain replacement, HM), contributions from wear, assembly and combustion are always constant over the years and over the countries. On the other hand, energy utilization contributions change over the countries and over the years due to the differences in production mixes, in distribution pathways and in overall local consumptions, which depend on local average temperatures. Poland in 2024 is characterized by a heavily fossil fuel-based electricity mix, and thus the share of the electricity utilization is almost three times the ones from Italy and Sweden. Going from 2024 to 2050, the gradual decarbonization of the production mixes affects overall emissions, reducing the weight of the energy utilization share. Moreover, the shift towards pipeline distribution with respect to truck trailers in 2050 allows for a huge drop in the emissions from this specific contribution, from more than 50% of overall pollutant emissions in 2024 to less than 20% in 2050. In addition, battery production has a huge impact on overall emissions, especially for BEVs, which are the worst-performing layout, regardless of the year and of the country, despite the expected reductions. For the sake of transparency, even though BEVs have already become a widespread solution all over the world, the largest export share in the battery manufacturing market for automotives is held by China [147,148,149]. Moreover, several studies have been carried out on the distribution of pollutants caused by international trades involving China [150,151,152], showing that deaths in the EU can be traced back to pollutants emitted in China and spread globally due to economic exchanges of goods. Production of goods is usually located in China not only due to its richness in raw materials but also owing to lower labor costs and less stringent environmental regulations [150]. Considering these aspects, it was assumed that all batteries shared the same associated manufacturing CO_add emissions, without distinguishing between each specific country of the EU27 where they are employed in BEVs. This could be seen as counter-intuitive, being that pollutant emissions have a local effect, in contrast to GHG emissions, which instead are impactful on a global scale. In reality, the scope of this study is not to look after individual EU27 countries’ interests, but to limit global pollution as a whole, to avoid an excessively NIMBY perspective: the strong localization of manufacturing, associated with a constantly rising demand from the EU, increases health burdens on China as well [151]. Thus, from both an ethical and a technical point of view, due to the large volumes of goods transported in such trades, it is not correct to assume that pollutant emissions generated from manufacturing in China will not have repercussion on the EU as well.

On the topic of energy storage shares, H₂ tanks have a much more limited impact. On the other hand, comparing the contributions of H2ICEs, electric motors and FCs on the powertrain share, the first has the lowest impact, being simply derived from an already well-established technology. As for electric motors, their contribution is larger than that of H2ICEs, but the most noticeable impact is represented by FCs. Despite the expected reductions in their production-related emissions, even in 2050 they are not able to reach H2ICEV and HH2ICEV powertrain shares.

4. Conclusions

In conclusion, this study provides a thorough analysis and side-by-side assessment of equivalent Human Toxicity Potential (HTP) emissions of four novel powertrain layouts (a Battery Electric Vehicle, BEV, a pure Hydrogen Internal Combustion Engine Vehicle, H2ICEV, a hybrid H2ICEV, HH2ICEV, and a Fuel Cell Vehicle, FCV) for urban buses, across all European Union 27 countries (EU27), over their whole lifecycle, in current and future frameworks (2024, 2030 and 2050), comparing different alternative scenarios for each, depending on the sensitivity to several parameters (curb weight, number of powertrain replacements and hydrogen production mix). BEVs show the highest CO_add emissions across all scenarios, while Hydrogen-Fueled Vehicles (HFVs) display lower CO_add emissions, similar among them. The findings underscore the following:

• BEV CO_add emissions are expected to noticeably drop over the years across all scenarios (to around half of 2024 values), mainly due to greener electricity mixes and reduced emissions from material extraction and manufacturing of batteries. Regardless, BEV emissions are vastly larger than those of HFVs (from up to four times in 2024 to up to double HFVs’ ones in 2050);
• BEVs’ and FCVs’ CO_add emissions are more negatively affected by an increase in powertrain replacement rather than by a curb weight increase (with BEVs showing the strongest dependency), due to the heavy impact of the raw material extraction and manufacturing phases. Conversely, H2ICEVs and HH2ICEVs show an opposite sensitivity, being that their manufacturing-related emissions are already very low but are less efficient than batteries and FCs, thus increasing the consumptions;
• Progressive transition to greener hydrogen production mixes over the years, coupled with a transition in H₂ distribution pathways, limits the improvement provided by 100 GH scenarios with respect to HM ones. In particular, in 2050, HM and 100 GH are so close to each other that, if combined with other conditions (i.e., doubling the powertrain replacement and/or increasing the curb weight), they always lead to higher CO_add emissions with respect to the baseline scenario;
• Looking at specific contributions to overall CO_add emissions in the baseline scenarios, batteries’ manufacturing has a huge impact. As a matter of fact, over all years and across all EU27 countries, despite forecasted reductions, the BEVs’ share related to energy storage alone is comparable to the overall emissions from the other HFVs;
• Emissions related to the H₂ distribution pathway heavily influence the overall pollutant emissions of HFVs, representing more than 50% of overall emissions in baseline scenarios, when considering truck trailers. Pipeline networks, on the other hand, allow us to noticeably reduce their impact.

These results highlight that current efforts must be intensified, especially for BEVs and for the H2 distribution network development, to further limit harmful emissions from the mobility sector. Moreover, they are crucial for stakeholders and policymakers in making informed decisions and prioritizing one solution over others. As a matter of fact, future developments should aim for an analysis that encompasses different aspects as well, such as Greenhouse Gas emissions and economic viability, to take a holistic approach and have a broader view on these kinds of novel technologies.